--- name: creative-testing description: Structure and read creative A/B tests across platforms — set up variants, decide when a result is real, and pick winners without fooling yourself on noise. Use when a user wants to test ads, compare creative variants, run an A/B test, or choose a winning ad. --- # Creative Testing Test to learn, then scale the winner. Most "winners" called early are noise. Reads do not require approval; launching variants spends money, so confirm before shipping. Confirm the account: `list_connected_accounts`. ## 1. Test one thing at a time Change one variable per test (hook, image, headline, offer, format) so the result is attributable. Testing five things at once tells you nothing about which one moved the number. ## 2. Set it up cleanly Write text variants with `create_ad_copy` and select existing approved visual assets with `list_creative_assets`. Keep the comparison controlled and give each variant enough data. Follow **launch** and **campaign-preflight** before deploying a test. ## 3. Wait for enough data - Judge on conversions and CPA where volume allows, on CTR only as an early proxy. - Do not call a winner off a few conversions or one day; small samples swing wildly. State the volume behind the call. - On Meta, respect the Learning Phase before comparing (see **platform-meta**). ## 4. Read and act Pull results per variant: the supported performance tools (`pull_google_ads_performance`, `pull_meta_ads_performance`, `pull_linkedin_ads_performance`, `pull_microsoft_ads_performance`, `pull_reddit_ads_performance`). Pick the winner on the metric that matters (usually CPA or ROAS, not CTR). Scale it with the **optimize** skill, retire the losers, and feed what you learned into the next test. ## Rules - One variable per test; enough data before a verdict. - Winner is decided on the conversion metric, not clicks, unless volume forces a proxy. - Report the variants, the volume, the winner, and the next test to run. Watch winners for fatigue with the **creative-fatigue-detector** skill.